U.S. patent application number 13/557641 was filed with the patent office on 2013-01-31 for providing social product recommendations.
This patent application is currently assigned to ALIBABA GROUP HOLDING LIMITED. The applicant listed for this patent is Shanshu Leng. Invention is credited to Shanshu Leng.
Application Number | 20130030950 13/557641 |
Document ID | / |
Family ID | 47575265 |
Filed Date | 2013-01-31 |
United States Patent
Application |
20130030950 |
Kind Code |
A1 |
Leng; Shanshu |
January 31, 2013 |
PROVIDING SOCIAL PRODUCT RECOMMENDATIONS
Abstract
Providing social product recommendations is disclosed,
including: determining product information of interest to a target
user; retrieving a plurality of product reviews associated with the
product information of interest, wherein the plurality of product
reviews is generated by a plurality of reviewer users; determining
evaluation values corresponding to the plurality of product
reviews; determining friendship dimension values between the
plurality of reviewer users and the target user; and determining a
recommendation value for a product associated with the plurality of
product reviews based on one or more evaluation values
corresponding to the product and weights associated with the one or
more evaluation values, wherein the weights are determined based at
least in part on friendship dimension values corresponding to those
of the plurality of reviewer users associated with those of the
plurality of product reviews associated with the product.
Inventors: |
Leng; Shanshu; (Hangzhou,
CN) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
Leng; Shanshu |
Hangzhou |
|
CN |
|
|
Assignee: |
ALIBABA GROUP HOLDING
LIMITED
George Town
KY
|
Family ID: |
47575265 |
Appl. No.: |
13/557641 |
Filed: |
July 25, 2012 |
Current U.S.
Class: |
705/26.7 |
Current CPC
Class: |
G06Q 30/02 20130101;
G06Q 50/01 20130101 |
Class at
Publication: |
705/26.7 |
International
Class: |
G06Q 30/00 20120101
G06Q030/00 |
Foreign Application Data
Date |
Code |
Application Number |
Jul 26, 2011 |
CN |
201110210210.X |
Claims
1. A system for providing social product recommendations,
comprising: one or more processors configured to: determine product
information of interest to a target user; retrieve a plurality of
product reviews associated with the product information of
interest, wherein the plurality of product reviews is generated by
a plurality of reviewer users; determine evaluation values
corresponding to the plurality of product reviews; determine
friendship dimension values between the plurality of reviewer users
and the target user; and determine a recommendation value for a
product associated with the plurality of product reviews based on
one or more evaluation values corresponding to the product and
weights associated with the one or more evaluation values, wherein
the weights are determined based at least in part on friendship
dimension values corresponding to those of the plurality of
reviewer users associated with those of the plurality of product
reviews associated with the product; and one or more memories
coupled to the one or more processors and configured to provide the
one or more processors with instructions.
2. The system of claim 1, wherein the one or more processors are
further configured to present the product with other products
associated with the plurality of product reviews based on the
product and the other products' recommendation values.
3. The system of claim 2, wherein the product and other products
are presented in a sequence determined by ranking their
recommendation values.
4. The system of claim 1, wherein product information of interest
to the target user includes information published by the target
user at a social media website, keywords of searches performed by
the target user, and/or information associated with webpages
browsed by the target user.
5. The system of claim 1, wherein evaluation values are determined
based on one or both of historical transaction information and
product ratings of their corresponding product reviews.
6. The system of claim 1, wherein the friendship dimension values
between the plurality of reviewer users and the target user are
determined based at least in part on pre-stored friendship
dimension information.
7. The system of claim 1, wherein the one or more processors are
further configured to sort the plurality of product reviews into
one or more groups, wherein each group includes those of the
plurality of product reviews associated with a unique product.
8. The system of claim 1, wherein the weights are also determined
at least in part on statuses of those of the plurality of reviewer
users associated with those of the plurality of product reviews
associated with the product.
9. The system of claim 1, wherein the recommendation value for the
product is determined as a weighted mean of the one or more
evaluation values corresponding to the product and the weights
assigned to the one or more evaluation values.
10. A method for providing social product recommendations,
comprising: determining product information of interest to a target
user; retrieving a plurality of product reviews associated with the
product information of interest, wherein the plurality of product
reviews is generated by a plurality of reviewer users; determining
evaluation values corresponding to the plurality of product
reviews; determining friendship dimension values between the
plurality of reviewer users and the target user; and determining a
recommendation value for a product associated with the plurality of
product reviews based on one or more evaluation values
corresponding to the product and weights associated with the one or
more evaluation values, wherein the weights are determined based at
least in part on friendship dimension values corresponding to those
of the plurality of reviewer users associated with those of the
plurality of product reviews associated with the product.
11. The method of claim 10, further comprising presenting the
product with other products associated with the plurality of
product reviews based on their recommendation values.
12. The method of claim 11, wherein the product and other products
are presented in a sequence determined by ranking their
recommendation values.
13. The method of claim 10, wherein product information of interest
to the target user includes information published by the target
user at a social media website, keywords of searches performed by
the target user, and/or information associated with webpages
browsed by the target user.
14. The method of claim 10, wherein evaluation values are
determined based on one or both of historical transaction
information and product ratings of their corresponding product
reviews.
15. The method of claim 10, wherein the friendship dimension values
between the plurality of reviewer users and the target user are
determined based at least in part on pre-stored friendship
dimension information.
16. The method of claim 10, further comprising sorting the
plurality of product reviews into one or more groups, wherein each
group includes those of the plurality of product reviews associated
with a unique product.
17. The method of claim 10, wherein the weights are also determined
at least in part on statuses of those of the plurality of reviewer
users associated with those of the plurality of product reviews
associated with the product.
18. The method of claim 10, wherein the recommendation value for
the product is determined as a weighted mean of the one or more
evaluation values corresponding to the product and the weights
assigned to the one or more evaluation values.
19. A computer program product for providing social product
recommendations, the computer program product being embodied in a
computer readable storage medium and comprising computer
instructions for: determining product information of interest to a
target user; retrieving a plurality of product reviews associated
with the product information of interest, wherein the plurality of
product reviews is generated by a plurality of reviewer users;
determining evaluation values corresponding to the plurality of
product reviews; determining friendship dimension values between
the plurality of reviewer users and the target user; and
determining a recommendation value for a product associated with
the plurality of product reviews based on one or more evaluation
values corresponding to the product and weights associated with the
one or more evaluation values, wherein the weights are determined
based at least in part on friendship dimension values corresponding
to those of the plurality of reviewer users associated with those
of the plurality of product reviews associated with the product.
Description
CROSS REFERENCE TO OTHER APPLICATIONS
[0001] This application claims priority to People's Republic of
China Patent Application No. 201110210210.X entitled A METHOD AND
EQUIPMENT OF RELEASING PRODUCT INFORMATION filed Jul. 26, 2011
which is incorporated herein by reference for all purposes.
FIELD OF THE INVENTION
[0002] The present invention relates to the field of computer
technology. In particular, it relates to the technique of
recommending product information.
BACKGROUND OF THE INVENTION
[0003] Online shopping has, thanks to its convenience and
flexibility, enjoyed steady growth and popularity. Via online
shopping, users can browse for and purchase products without having
to leave their homes. Moreover, users may be able to make more
informed purchases through perusing the great abundance of product
information online and also better perform comparisons between
different products before ultimately making a transaction.
[0004] However, due to the enormous amounts of product information
that is available online, online shoppers may need to sift through
a lot of information before they find relevant content. Therefore,
it would be desirable to determine the product information that
would enable online shoppers to quickly find the products that best
meet their interests.
[0005] Conventionally, when product information is presented and/or
recommended to a user at an online shopping platform, products are
generally ranked based on a determined metric based on one or more
of the following, for example: product characteristics, merchant
trustworthiness, product price, and merchant address. For example,
in response to a user's keyword-based search for products, the
online shopping platform will rank the search results according to
the user's requirements on product price, product characteristics,
merchant address, and/or merchant trustworthiness, for example, and
display the ranked search results to the user. However, the
conventional technique of returning search results for the user may
not generate the search results desirable for (or search results
ranked in a manner suitable to the interests of) the user, which
would lead the user to submit differently constructed search
requests that describe the same product or same type of products
until desirable search results are received. Repeated resubmission
of search requests may be inefficient and also frustrating for the
user.
BRIEF DESCRIPTION OF THE DRAWINGS
[0006] Various embodiments of the invention are disclosed in the
following detailed description and the accompanying drawings.
[0007] FIG. 1 is a diagram showing an embodiment of a system for
social product recommendations.
[0008] FIG. 2 is a flow diagram showing an embodiment of a process
for providing social product recommendations.
[0009] FIGS. 3A, 3B, and 3C illustrate examples of determining the
friendship dimension values for users A and B using portions of
pre-stored representations.
[0010] FIG. 4 is a flow diagram showing an embodiment of an example
for providing social product recommendations.
[0011] FIG. 5 is a diagram showing an embodiment of a system for
providing social product recommendations.
DETAILED DESCRIPTION
[0012] The invention can be implemented in numerous ways, including
as a process; an apparatus; a system; a composition of matter; a
computer program product embodied on a computer readable storage
medium; and/or a processor, such as a processor configured to
execute instructions stored on and/or provided by a memory coupled
to the processor. In this specification, these implementations, or
any other form that the invention may take, may be referred to as
techniques. In general, the order of the steps of disclosed
processes may be altered within the scope of the invention. Unless
stated otherwise, a component such as a processor or a memory
described as being configured to perform a task may be implemented
as a general component that is temporarily configured to perform
the task at a given time or a specific component that is
manufactured to perform the task. As used herein, the term
`processor` refers to one or more devices, circuits, and/or
processing cores configured to process data, such as computer
program instructions.
[0013] A detailed description of one or more embodiments of the
invention is provided below along with accompanying figures that
illustrate the principles of the invention. The invention is
described in connection with such embodiments, but the invention is
not limited to any embodiment. The scope of the invention is
limited only by the claims and the invention encompasses numerous
alternatives, modifications and equivalents. Numerous specific
details are set forth in the following description in order to
provide a thorough understanding of the invention. These details
are provided for the purpose of example and the invention may be
practiced according to the claims without some or all of these
specific details. For the purpose of clarity, technical material
that is known in the technical fields related to the invention has
not been described in detail so that the invention is not
unnecessarily obscured.
[0014] Providing social product recommendations for an online
shopping platform is described herein. In various embodiments, a
target user may indicate product information of interest. Then
product reviews are received from users other than the target user
for the products associated with the product information of
interest. Evaluation values for the product reviews associated with
the product information of interest may be determined based at
least in part on portions of the received product reviews. Also, a
friendship dimension value is determined between the target user
and each other user that submitted a product review. In some
embodiments, a recommendation value is determined for each unique
product described in the product reviews based at least in part on
the evaluation values determined for product reviews associated
with that product and the friendship dimension values associated
with the reviewer users that submitted the reviews for that
product. In some embodiments, products are then recommended to the
target user based on a ranking determined based on their respective
recommendation values.
[0015] FIG. 1 is a diagram showing an embodiment of a system for
social product recommendations. In the example, system 100 includes
device 102, network 104, and server 106. Network 104 may include
high speed and/or telecommunications networks.
[0016] Device 102 is configured to communicate with server 106 over
network 104. While device 102 is shown to be a laptop, other
examples of device 102 may be a desktop computer, a tablet, a
smartphone, a mobile device, or any other type of computing device.
Device 102 is installed with a web browser application that enables
a target user, a user who is interested in potentially purchasing
items at an online shopping platform, to indicate product
information of interest by sharing information at a social media
website, through browsing webpages, and/or performing keyword-based
searches for products of interest. Such product information of
interest for the target user may be determined by server 106.
[0017] Server 106 is configured to support an online shopping
platform. For example, the online shopping platform may be
accessible at a particular uniform resource locator (URL) with a
web browser and may enable products to be sold and bought by
individual users. Using determined product information of interest
to the target user, server 106 is configured to retrieve related
product reviews submitted by other users. Server 106 is configured
to determine an evaluation value for each product review using at
least a portion of the product review. Server 106 is also
configured to determine a friendship dimension value between the
target user and each reviewer user that submitted a product review,
where the friendship dimension value indicates the closeness of the
friendship between the two users at the online shopping platform.
Generally, the greater the friendship dimension value is, the
closer friends the two users are, and therefore, the more influence
the reviewer user's product review exerts on the degree to which
the reviewed product(s) will be recommended to the user. As such,
server 106 is configured to use at least the determined friendship
dimension values for reviewer users and the target user to
determine the recommendation values associated with the reviewed
products, which, in some embodiments, will be recommended to the
user in a ranking based on their respective recommendation
values.
[0018] FIG. 2 is a flow diagram showing an embodiment of a process
for providing social product recommendations. In some embodiments,
process 200 may be implemented at system 100.
[0019] At 202, product information of interest to a target user is
determined. In some embodiments, the target user is a user for whom
product recommendations are to be determined. Production
information of interest may include information that describes
products (e.g., specific products or just types of products) that
the target user may be interested in purchasing. In some
embodiments, product information of interest to the target user may
be determined from the online operations of the user. Product
information of interest to the target user may be determined using
one or more techniques. In one example, the product information of
interest to the target user may be determined from stored browsing
history associated with the target user. For instance, product
information of interest may include the features common to the
products whose sale webpages were browsed by the target user more
frequently. In another example, the product information of interest
to the target user may be determined from a keyword-based search
that is being performed by the target user. For instance, the
product information of interest may include the keywords that are
input into the search input box. In yet another example, the
product information of interest to the target user may be
determined from information published (e.g., shared) by the target
user on a webpage, blog, and/or social networking website. For
instance, the product information of interest to the target user
may include information posted by the target user at a microblog
(e.g., if the target user posts "getting ready to replace my mobile
phone," then "mobile phone" may be used as the product information
of interest).
[0020] In some embodiments, the determined product information of
interest is used to determine one or more product categories in
which the target user has an interest. For example, such product
categories may be determined by performing a keyword match between
the determined product information of interest to the target user
and keywords associated with one or more predetermined product
categories available at an online shopping platform. In another
example, such product categories may be determined based on the
target user's selection(s) of product categories.
[0021] At 204, a plurality of product reviews associated with the
product information of interest is retrieved, wherein the plurality
of product reviews is generated by a plurality of reviewer users.
In some embodiments, product reviews submitted for various products
at the online shopping platform are stored. Each product review may
be submitted by a user associated with an account at the online
shopping platform. In some embodiments, each user is assigned a
user ID. As will be discussed below, product reviews associated
with the product information of interest may be retrieved by
searching through stored product reviews, by receiving product
reviews submitted in response to a publication by the target user,
and/or by tracking bookmarking actions.
[0022] In some embodiments, a search may be performed for stored
product reviews (submitted by users other than the target user) on
products that are associated with the product information of
interest to the target user as determined in 202. In some
embodiments, a user that has submitted a product review is referred
to as a reviewer user. For example, each product review may include
one or more of: one or more identifiers of the reviewed product, a
rating on a certain scale, an image, text description, the user ID
of the reviewer user of the product review, and/or historical
transaction information regarding the reviewer user and the
reviewed product (e.g., whether the reviewer user has successfully
purchased the product and whether the reviewer user has returned
the product). In some embodiments, stored product reviews submitted
by various reviewer users are searched to determine those product
reviews that are authored by users who are associated with a user
ID other than the user ID of the target user and that are also
associated with products in the determined product categories in
which the target user has an interest. In some embodiments, only
product reviews received within a predetermined time period (e.g.,
the last month) are searched. For example, if the target user had
posted on a social network website "getting ready to buy a new
mobile phone," then product reviews submitted by other users within
the last month for various types of mobile phones are searched for
and retrieved.
[0023] In some embodiments, product reviews are submitted by
reviewer users in response to the target user's publication of
production information of interest. For example, if the target user
had posted on a social network website "getting ready to buy a new
mobile phone," then other users may reply to the post with product
reviews of products associated with "mobile phone." In some
embodiments, the target user may be presented with fields and/or
selections additional to those a user normally uses to post
information at a social network website. Examples of such
additional fields and/or selections presented for the target user
may include product category and product model. In some
embodiments, for such product review replies, additional fields or
selections may be presented to replying users to fill out that may
include, for example, product category, product model, key
attributes (e.g., color, dimensions, performance parameters), and
price.
[0024] In some embodiments, product reviews are implied through
bookmarking actions of products associated with the product
information of interest. For example, the historical bookmarking
actions of various users (who are reviewing the products by virtue
of bookmarking them) are recorded.
[0025] Retrieved product reviews may be associated with one or more
unique products. For example, among the 10 retrieved product
reviews, 5 may be for one unique product, 3 may be for a second
unique product, and 2 may be for a third unique product.
[0026] At 206, evaluation values corresponding to the plurality of
product reviews are determined.
[0027] In some embodiments, an evaluation value is determined for
each retrieved product review. The range of evaluation values can
be set based on any appropriate scheme. In an example scheme, the
greater the evaluation value, the more favorable the product review
is of the product. In various embodiments, determining an
evaluation value for a product review includes mapping at least a
portion of the information (e.g., the historical transaction
information and/or rating) in the product review into a numerical
value on the evaluation value range of the chosen scheme. For
example, if the scheme involved assigning evaluation values for the
two historical transaction information options of "successful
transaction" and "product returned," then a different evaluation
value can be assigned to each option. For instance, the evaluation
value for "successful transaction" may be set as 3 and the
evaluation value for "product returned" may be set as -1. In
another example, if the scheme involved assigning evaluation values
for product ratings, then a different evaluation value may be
assigned to each product rating that includes "Very good," "Good,"
"Average," "Poor," and "Very poor." For instance, the evaluation
value for "Very good" may be set at 3, the evaluation value for
"Good" may be set at 2, the evaluation value for "Average" may be
set at 1, the evaluation value for "Poor" may be set as -1, and the
evaluation value for "Very poor" may be set as -2. In yet another
example, if the scheme involved assigning evaluation values for a
combination of product ratings and historical transactional
information, then a different evaluation value may be assigned to
each different combination of historical transaction options and
product ratings. For instance, to provide example evaluation values
for just a few of such possible combinations, the evaluation value
for the combination of "successful transaction" and a product
rating of "Very good" may be set as 5, the evaluation value for the
combination of "product returned" and a product rating of "Good"
may be set as 2, and the evaluation value for the combination of
"successful transaction" and a product rating of "Poor" may be set
as -2.
[0028] In some embodiments, in addition or alternative to the
historical transaction information and product ratings of product
reviews, product reviews that comprise bookmarking records
associated with the users can also serve as a basis for determining
evaluation values for a product. For example, if the target user
had performed a search with the keywords "flip phone," then the
evaluation values for these products associated with "flip phone"
are determined by, for example, setting an evaluation value for
each such product as the total number of times that the product has
been bookmarked by one or more users other than the target
user.
[0029] At 208, friendship dimension values between the plurality of
reviewer users and the target user are determined. In some
embodiments, a friendship dimension value is determined between
each reviewer user whose product review has been retrieved and the
target user. In some embodiments, the online shopping platform
includes users at the platform to form platform-recognized
relationships with each other. For example, one such
platform-recognized relationship may be a friendship relationship
and two users that are friends at the platform may be thought of as
directly linked in a social graph that represents all the
platform-recognized relationships of the platform. In some
embodiments, a "friendship dimension value" refers to a numerical
value that represents the closeness in friendship between two users
(e.g., two user IDs) at the platform.
[0030] The following is one example technique by which a friendship
dimension value may be determined between a reviewer user and a
target user:
[0031] A social graph or other representation of friendship
relationships among users of an online shopping platform is
pre-stored in a friendship dimension database. For example, the
social graph may be stored as one or more tables that indicate the
friendship relationships between each user and every other user
with whom the user is friends. In a visual representation of this
pre-stored information, each user at the platform may be
represented as a node associated with the user's respective user ID
and each friendship relationship between two users may be
represented by a link between the nodes corresponding to these two
users. Each node in the representation may be linked to zero or
more other nodes (i.e., each user at the platform may have zero or
more friends at the platform). Friendship dimensions between the
reviewer users (i.e., the users associated with the product reviews
that have been retrieved) and the target user may be determined by
identifying the nodes corresponding to the reviewer users and the
target user, as well as the links between such nodes. Each reviewer
user may or may not be directly linked to the target user. For
example, if the reviewer user and the target user were friends
(e.g., as indicated by the tables of pre-stored information), then
a link exists between the nodes of this reviewer user and the
target user. However, if the reviewer user were not directly
friends with the target user, then the reviewer user may be
indirectly linked to the target user by virtue of shared friends.
For example, shared friends may be the nodes between a first user
and a second user (e.g., the reviewer user and the target user) and
that each of such nodes may be reached through traversing the
continuous links from either the first user to the second user or
the second user to the first user. In one example, a friendship
dimension value between a reviewer user and the target user may be
determined as the minimum number of links in between that reviewer
user and the target user in the pre-stored representation. FIGS.
3A, 3B, and 3C illustrate examples of determining the friendship
dimension values for users A and B using portions of pre-stored
representations. FIG. 3A shows two users, A and B, who are directly
friends with each other and hence a link connects the two users.
Thus, the friendship dimension value between users A and B in this
example is one. FIG. 3B shows four users, A, B, C, and D, in which
A is not directly friends with B but is friends with C, who is in
turn friends with D, who is in turn friends with B. Thus, the
friendship dimension value between A and B in this example is three
because there are three links in between users A and B. FIG. 3C
shows five users, A, B, C, D, and E, in which A is not directly
friends with B. However, A may be linked to B through just D or
through C, D, and E. Because two links (the number of links from A
to D and from D to B) is the minimum number of links between users
A and B (as opposed to the four links between A and C, C and D, D
and E, and E and B), the friendship dimension value between A and B
in this example is two.
[0032] Referring back to FIG. 2, as shown above, the smaller the
friendship dimension value between the reviewer user and the target
user, the more close the users are at the online shopping platform
and potentially, the more influential the reviewer user's product
reviews are to the target user in terms of affecting product
recommendations for the target user.
[0033] At 210, a recommendation value for a product associated with
the plurality of product reviews is determined based on one or more
evaluation values corresponding to the product and weights
associated with the one or more evaluation values, wherein the
weights are determined based at least in part on friendship
dimension values corresponding to those of the plurality of
reviewer users associated with those of the plurality of product
reviews associated with the product.
[0034] In some embodiments, first, a weight is determined for each
evaluation value determined for a product review based at least in
part on a friendship dimension value associated with the reviewer
user that authored the product review. In some embodiments, in
addition to the friendship dimension value associated with a
reviewer user, a weight for an evaluation value and/or the
evaluation value itself may be adjusted based on a particular
status associated with the reviewer user that authored that product
review. Then, once weights have been determined for the evaluation
values, weighted evaluations associated with the same product
described in the product reviews (e.g., as identified by the
identifiers of the reviewed products) are grouped together and used
to determine the recommendation value for that product. In various
embodiments, a recommendation value indicates a metric of product
recommendations for the target user. Generally, the greater the
magnitude of the recommendation value of a product, the more the
value indicates to the target user that the product is favorably
reviewed by other users (e.g., friends and shared friends of the
target user) at the online shopping platform.
[0035] As mentioned above, before determining the recommendation
value for each product associated with the retrieved product
reviews, each evaluation value determined for a product review is
weighted by a weight determined based at least in part on the
friendship dimension value of the user ID of the reviewer user that
authored that product review and/or the reviewer's status (e.g.,
buyer, seller, or operator.)
[0036] For example, corresponding weights may be determined for
evaluation values based on friendship dimension values using a
predetermined scheme. In general, the greater the weight assigned,
the greater influence the reviewer user is assigned to have on the
product recommendations for the target user. In some example
scheme, the weight associated with a friendship dimension value of
one (i.e., the target user and the reviewer user are directly
linked) is set to 6, the weight associated with a friendship
dimension value of two (i.e., the target user and the reviewer user
are linked via one shared friend) is set to 5, the weight
associated with a friendship dimension value of three (i.e., the
target user and the reviewer user are linked via two shared
friends) is set to 4, the weight associated with a friendship
dimension value of four (i.e., the target user and the reviewer
user are linked via three shared friends) is set to 3, the weight
associated with a friendship dimension value of five (i.e., the
target user and the reviewer user are linked via four shared
friends) is set to 2, the weight associated with a friendship
dimension value of six (i.e., the target user and the reviewer user
are linked via five shared friends) is set to 1, and the weight
associated with a friendship dimension value greater than six
(i.e., the target user and the reviewer user are linked via more
than five shared friends) is also set to 1 (because it is assumed
that two users associated with a friendship dimension value of
greater than six are not very close already and as such, their
relationship need not be further distinguished as the friendship
dimension value increases further).
[0037] As mentioned above, the weights assigned to evaluation
values may be adjusted based on a certain status of the reviewer
user. For example, the status may include whether the reviewer user
is a buyer, a seller, or an operator associated with the online
shopping platform. For example, a seller user may comprise a user
who sells at the online shopping platform the very product for
which he submitted a product review, an operator user is a user
that is an employee or otherwise affiliated with the online
shopping platform, and a buyer user is neither a seller nor an
operator. For example, the weights corresponding to buyer-status
user IDs of friendship dimension values one through six may be,
respectively, 12, 10, 8, 6, 4 and 2, while the weights
corresponding to seller-status user IDs of friendship dimension
values one through six may be, respectively, 3, 2.5, 2, 1.5, 1 and
0.5. In this example, the weight corresponding to buyer-status user
IDs is greater than the weight corresponding to seller-status user
IDs of the same friendship dimension value. For example, the weight
set for every friendship dimension value one through six
corresponding to a user ID having online shopping platform operator
status may be 10. In other words, in this example, the same weight
is set for every friendship dimension value one through six if the
reviewer user has an operator status (unlike for users with seller
or buyer statuses, where varying weights are set for different
friendship dimension values).
[0038] In addition, in some embodiments, it is also possible to set
special weights for users having a platform-recognized special
relationship with the target user. For example, a special
relationship may be different from a relationship described by the
friendship dimension value. For example, a weight of 8 may be
assigned to product reviews and their corresponding evaluation
values submitted by a close friend or relative of the target
user.
[0039] In some cases, where the reviewer user has a seller status,
his or her product review might be biased towards making more sales
of the reviewed product (e.g., the seller's evaluation of the
product may be biased towards being very favorable). Thus, in some
embodiments, the bias of such seller user provided product reviews
are reduced by attributing a weight of a lower magnitude to such
reviews and thus, their respective evaluation values are
accordingly attenuated. Furthermore, the evaluation value of a
seller reviewer user need only be considered for whether it is a
positive value or a negative value. A specific example could be as
follows: Prior to calculating the recommendation value for a
product, adjust all positive evaluation values among evaluation
values whose corresponding user status is seller to a standardized
positive value (e.g., set all positive evaluation values for
sellers to 1), and adjust (e.g., set) all negative evaluation
values among evaluation values whose corresponding user status is
seller to a standardized negative value (e.g., set all negative
evaluation values for sellers to -0.8). By adjusting evaluation
values associated with product reviews by seller reviewer users to
standardized positive or negative values, the influence of seller
users are more or less only viewed as positive or negative, without
different degrees of positivity or negativity.
[0040] Additionally, in some embodiments, when a reviewer user is
determined to be of a seller status, the weight assigned to the
evaluation value(s) associated with that reviewer user may be
determined based on a measure of seller credibility (or
creditworthiness) associated with that reviewer user at the
platform. For example, the online shopping platform may establish a
credibility for each seller at the platform based at least in part
on buyers' reviews of the seller's sales. For example, the higher
the seller user's credibility is, the higher the weight that is
assigned to an evaluation value associated with a product review
authored by that seller user.
[0041] As mentioned above, once weights for evaluation values have
been determined, then recommendation values may be determined for
each unique product described by the product reviews. In some
embodiments, because the retrieved product reviews may include
multiple product reviews for the same product, the retrieved
product reviews may be sorted into groups, where product reviews
describing the same product (as identified by the product's
identifier) will be sorted into the same group. As a result, each
group of product reviews will be associated with one unique
product. Then the evaluation values and their respective weights of
all the product reviews in a group are used to determine a
recommendation value for the product associated with that group of
product reviews.
[0042] In some embodiments, a recommendation value for a particular
product is determined to be the weighted mean of all the weighted
evaluation values associated with that product. A weighted mean may
be determined as the sum of the products between each evaluation
value and its respective weight divided by the sum of all the
weights. For example, if for product A, the first associated
evaluation value is 5 and the respective weight is 2, and the
second associated evaluation value is 6 and the respective weight
is 4, then the weighted mean (recommendation value) for product A
will be (5*2+6*4)/(2+4)=34/6=5.67.
[0043] In some embodiments, a determined recommendation value for a
particular product may be adjusted as follows: If all the reviewer
users who have a friendship dimension value of one with the target
user (i.e., users who are friends with the target user) provide the
highest evaluation for a certain product, then the recommendation
value of the product may be adjusted to a higher value (or the
highest value among all the products) so that the product will be
preferentially recommended to the user. Or, if the majority of
reviewer users who have a friendship dimension value of one with
the target user (i.e., users who are friends with the target user)
provide the lowest evaluation for a product, then the
recommendation value of the product may be adjusted to zero so that
the product will not be recommended (at least among the earlier set
of recommended products) to the user. By adjusting the
recommendation value determined for a target user based on the
product reviews of the target user's friends, then the
recommendation values for the products collectively preferred or
disfavored by the friends will reflect such opinion.
[0044] At 212, the product is presented with other products
associated with the plurality of product reviews based on the
product and the other products' respective recommendation values.
In some embodiments, the unique products described by the retrieved
product reviews are ranked based on the product and the other
products' respective recommendation values and presented to the
target user (e.g., at a webpage) as a list based on the determined
ranking. Products associated with higher recommendation values are
ranked higher and indicate to the target user that such products
may be of more interest to him or her. For example, the
presentation of each product may include information related to
that product (e.g., product category, product model, color, price)
and a link to a webpage associated with selling that product. In
the event there are products tied with the same recommendation
value, then such products may be presented in a random sequence
relative to each other. In some embodiments, the products are not
ranked by the product and the other products' respective
recommendation value but rather, are displayed as a list and each
product is displayed with its respective recommendation value.
Because the products' respective recommendation values were
determined based on the target user's friendship dimension values
with users that have submitted reviews, the degree to which a
product is recommended to the target user (i.e., the magnitude of
the recommendation value) reflects not only the reviewer's opinion
of the product but also the reviewer's closeness of friendship
(i.e., based on the friendship dimension value) to the target user
at the online shopping platform.
[0045] FIG. 4 is a flow diagram showing an embodiment of an example
for providing social product recommendations. In some embodiments,
process 400 may be implemented at system 100. In some embodiments,
process 200 may be implemented using process 400.
[0046] At 402, product information of interest published by a
target user is determined. In some embodiments, product information
of interest to the target user includes at least keywords and/or
product categories. In some embodiments, publishing may include
sharing, posting, uploading, and/or updating at a blog, a website,
or a social networking website, for example. For example, the user
may publish the following information on a microblog: "getting
ready to replace my mobile phone," of which the keyword would be
"mobile phone."
[0047] At 404, a plurality of product reviews generated in response
to the product information of interest is received. Other users who
may have seen the product information of interest that was
published by the target user may want to reply to the publication
with some related product reviews. Returning to the previous
example, in response to the target user's post related to "mobile
phone," other users may reply to the post with product reviews on
mobile phones. Such responses may include one or more of: one or
more identifiers of the reviewed product, a rating on a certain
scale, an image, text description, the user ID of the reviewer user
of the product review, and/or historical transaction information
regarding the reviewer user and the reviewed product (e.g., whether
the reviewer user has successfully purchased the product and
whether the reviewer user has returned the product).
[0048] At 406, it is determined if each reviewer user associated
with the plurality of product reviews is associated with a seller
status. An example of a status other than seller is buyer. If not a
seller, control passes to 408, 410, and then to 416. Otherwise,
control passes to 412, 414, and then to 416. The determination at
406 is performed to treat the product reviews of users who are
sellers differently than product reviews of users who are not
sellers (e.g., buyers).
[0049] At 408, an evaluation value for a respective product review
associated with the reviewer user and a friendship dimension value
between the reviewer user and the target user are determined. For
example, a friendship dimension value between a reviewer user and
the target user may be determined using pre-stored friendship
dimension information.
[0050] At 410, a weight associated with the product review is
determined based at least in part on the friendship dimension value
and the status of the reviewer user.
[0051] At 412, an evaluation value for respective the product
review associated with the reviewer user and a friendship dimension
value between the reviewer user and the target user are determined,
and in the event the evaluation value is greater than 0, the
evaluation value is adjusted to 0 and in the event that the
evaluation value is less than 0, the evaluation value is adjusted
to -1. Because the reviewer user is a seller and may be interested
in biasing the product review, the evaluation value of the product
review is standardized to reduce the effect of the possible
bias.
[0052] At 414, a weight associated with product reviews is
determined based at least in part on the friendship dimension
value, the status of the reviewer user, and a measure of seller
credibility associated with the reviewer user.
[0053] At 416, a recommendation value is determined for a product
associated with the plurality of product reviews based on an
evaluation value and a respective weight corresponding to the
product. For example, a recommendation value is determined for each
unique product described among the product reviews based on the
evaluation values and respective weights determined for the product
reviews for that product.
[0054] At 418, the product is ranked among with other products
associated with the plurality of product reviews based on the
product and the other products' respective recommendation values.
For example, the list of products based on the products' respective
recommendation values may be presented to the target user. In some
embodiments, the products with the higher recommendation values are
presented earlier.
[0055] At 420, a selection associated with a product included in
the ranking is received. For example, the target user may select a
product displayed in the ranked list, in part due to the manner in
which the product was ranked and the selected product's
recommendation value, which was influenced the most by his closer
friends.
[0056] FIG. 5 is a diagram showing an embodiment of a system for
providing social product recommendations. System 500 includes first
acquisition module 310, second acquisition module 320,
determination module 330, and display module 340.
[0057] The modules can be implemented as software components
executing on one or more processors, as hardware such as
programmable logic devices and/or Application Specific Integrated
Circuits designed to perform certain functions, or a combination
thereof. In some embodiments, the modules can be embodied by a form
of software products which can be stored in a nonvolatile storage
medium (such as optical disk, flash storage device, mobile hard
disk, etc.), including a number of instructions for making a
computer device (such as personal computers, servers, network
equipment, etc.) implement the methods described in the embodiments
of the present invention. The modules may be implemented on a
single device or distributed across multiple devices.
[0058] In some embodiments, first acquisition module 310 is
configured to determine product information of interest to a target
user. In some embodiments, first acquisition module 310 is also
configured to retrieve product reviews associated with the product
information of interest to the target user.
[0059] In some embodiments, second acquisition module 320 is
configured to determine evaluation values for the product reviews
and the friendship dimension values between the reviewer users that
submitted the product reviews and the target user.
[0060] In some embodiments, determination module 330 is configured
to determine recommendation values for the products described by
the product reviews based on the evaluation values and friendship
dimension values.
[0061] In some embodiments, display module 340 is configured to
present information associated with the products, where the
products are ranked based on their respective recommendation values
or at least displayed with their respective recommendation
values.
[0062] In some embodiments, first acquisition module 310 is further
configured to determine the keywords of searches performed by the
target user or the browsing history information of the target user
or information published by the target user on a social media
website.
[0063] In some embodiments, second acquisition module 320 is
further configured to determine, using a friendship dimension
database that includes pre-stored information on friendships among
user IDs at the online shopping platform, the user IDs of reviewer
users corresponding to the evaluation values and the user ID of the
target user, and to determine the minimum number of links between
the two users as the friendship dimension value between the
reviewer user and the target user.
[0064] In some embodiments, second acquisition module 320 is
further configured to determine the historical transaction
information and/or product ratings of products among the product
reviews and to determine the evaluation value of each product
review at least in part using the historical transaction
information and/or product rating associated with the product
review.
[0065] In some embodiments, determination module 330 is further
configured to determine a weight corresponding to each evaluation
value based at least in part on the friendship dimension value of
the reviewer user associated with the evaluation value and a status
(e.g., buyer, seller, or an operator at the online shopping
platform) associated with the reviewer user. In some embodiments,
in the event that the status of a reviewer user was a seller,
determination module 330 is configured to also take the user's
seller credibility at the platform into account in determining the
weight for a evaluation value associated with the user's product
reviews. For example, determination module 330 is configured to
adjust all positive evaluation values associated with a seller
reviewer user into a standardized positive value and all negative
evaluation values associated with the seller reviewer user into a
standardized negative value. In some embodiments, determination
module 330 is configured to sort product reviews associated with
the same product into a group and use the evaluation values
determined for those product reviews and the evaluation values'
respective weights to determine a recommendation value for that
product. In some embodiments, the recommendation value is
determined to be the weighted mean of the evaluation values and
their respective weights.
[0066] In some embodiments, display module 340 is further
configured to rank the products based on the magnitudes of their
respective recommendation values and present products in the
sequence of the ranking.
[0067] Persons skilled in the art can understand that the described
modules may be implemented across distributed devices or within a
single device. The modules may be combined into a single module, or
they can be further divided into several sub-modules.
[0068] The description above is only a specific means of
implementing the present application. It should be pointed out that
persons with ordinary skill in the art can, without departing from
the principles of the present application, also produce a number of
improvements and embellishments, and that such improvements and
embellishments should also be regarded as falling within the scope
of protection of the present application.
[0069] Although the foregoing embodiments have been described in
some detail for purposes of clarity of understanding, the invention
is not limited to the details provided. There are many alternative
ways of implementing the invention. The disclosed embodiments are
illustrative and not restrictive.
* * * * *